Title: Economics of AI Safety: Autonomy, R&D, and Optimal Incentives

Speaker: Chen Yongmin, Provost Professor of Economics, University of Colorado Boulder.

Abstract: Artificial intelligence generates enormous benefits but also creates novel safety risks, including extreme risks for which full liability is infeasible. This talk presents an economic framework in which firms choose AI autonomy and invest in either "balanced" or "targeted" safety R&D. A central finding is that optimal liability for ordinary AI harm should be punitive—exceeding actual harm—to curb excessive autonomy and mitigate investment distortions. The talk further examines the conditions under which ex ante regulation dominates ex post liability. It also discusses how open-sourcing, market competition, and global rivalry alter AI safety incentives.

Date & time: 22 May 2026, 10:00 - 11:30

Venue: B321, Zhixin Building, Central Campus, Shandong University